| 1 | // Copyright (c) 2012, the Dart project authors. Please see the AUTHORS file |
| 2 | // for details. All rights reserved. Use of this source code is governed by a |
| 3 | // BSD-style license that can be found in the LICENSE file. |
| 4 | |
| 5 | #include "platform/assert.h" |
| 6 | #include "vm/bootstrap_natives.h" |
| 7 | #include "vm/canonical_tables.h" |
| 8 | #include "vm/exceptions.h" |
| 9 | #include "vm/native_entry.h" |
| 10 | #include "vm/object.h" |
| 11 | #include "vm/object_store.h" |
| 12 | #include "vm/regexp_assembler_bytecode.h" |
| 13 | #include "vm/regexp_parser.h" |
| 14 | #include "vm/reusable_handles.h" |
| 15 | #include "vm/symbols.h" |
| 16 | #include "vm/thread.h" |
| 17 | |
| 18 | #if !defined(DART_PRECOMPILED_RUNTIME) |
| 19 | #include "vm/regexp_assembler_ir.h" |
| 20 | #endif // !defined(DART_PRECOMPILED_RUNTIME) |
| 21 | |
| 22 | namespace dart { |
| 23 | |
| 24 | DEFINE_NATIVE_ENTRY(RegExp_factory, 0, 6) { |
| 25 | ASSERT( |
| 26 | TypeArguments::CheckedHandle(zone, arguments->NativeArgAt(0)).IsNull()); |
| 27 | GET_NON_NULL_NATIVE_ARGUMENT(String, pattern, arguments->NativeArgAt(1)); |
| 28 | |
| 29 | bool multi_line = arguments->NativeArgAt(index: 2) == Bool::True().ptr(); |
| 30 | bool ignore_case = arguments->NativeArgAt(index: 3) != Bool::True().ptr(); |
| 31 | bool unicode = arguments->NativeArgAt(index: 4) == Bool::True().ptr(); |
| 32 | bool dot_all = arguments->NativeArgAt(index: 5) == Bool::True().ptr(); |
| 33 | |
| 34 | RegExpFlags flags; |
| 35 | flags.SetGlobal(); // All dart regexps are global. |
| 36 | if (ignore_case) flags.SetIgnoreCase(); |
| 37 | if (multi_line) flags.SetMultiLine(); |
| 38 | if (unicode) flags.SetUnicode(); |
| 39 | if (dot_all) flags.SetDotAll(); |
| 40 | |
| 41 | RegExpKey lookup_key(pattern, flags); |
| 42 | RegExp& regexp = RegExp::Handle(zone: thread->zone()); |
| 43 | { |
| 44 | REUSABLE_OBJECT_HANDLESCOPE(thread); |
| 45 | REUSABLE_SMI_HANDLESCOPE(thread); |
| 46 | REUSABLE_WEAK_ARRAY_HANDLESCOPE(thread); |
| 47 | Object& key = thread->ObjectHandle(); |
| 48 | Smi& value = thread->SmiHandle(); |
| 49 | WeakArray& data = thread->WeakArrayHandle(); |
| 50 | data = thread->isolate_group()->object_store()->regexp_table(); |
| 51 | CanonicalRegExpSet table(&key, &value, &data); |
| 52 | regexp ^= table.GetOrNull(key: lookup_key); |
| 53 | table.Release(); |
| 54 | if (!regexp.IsNull()) { |
| 55 | return regexp.ptr(); |
| 56 | } |
| 57 | } |
| 58 | |
| 59 | // Parse the pattern once in order to throw any format exceptions within |
| 60 | // the factory constructor. It is parsed again upon compilation. |
| 61 | RegExpCompileData compileData; |
| 62 | // Throws an exception on parsing failure. |
| 63 | RegExpParser::ParseRegExp(input: pattern, regexp_flags: flags, result: &compileData); |
| 64 | |
| 65 | { |
| 66 | RegExpKey lookup_symbol_key(String::Handle(ptr: Symbols::New(thread, str: pattern)), |
| 67 | flags); |
| 68 | SafepointMutexLocker ml(thread->isolate_group()->symbols_mutex()); |
| 69 | CanonicalRegExpSet table( |
| 70 | thread->zone(), |
| 71 | thread->isolate_group()->object_store()->regexp_table()); |
| 72 | regexp ^= table.InsertNewOrGet(key: lookup_symbol_key); |
| 73 | thread->isolate_group()->object_store()->set_regexp_table(table.Release()); |
| 74 | } |
| 75 | |
| 76 | ASSERT(regexp.flags() == flags); |
| 77 | return regexp.ptr(); |
| 78 | } |
| 79 | |
| 80 | DEFINE_NATIVE_ENTRY(RegExp_getPattern, 0, 1) { |
| 81 | const RegExp& regexp = RegExp::CheckedHandle(zone, ptr: arguments->NativeArgAt(index: 0)); |
| 82 | ASSERT(!regexp.IsNull()); |
| 83 | return regexp.pattern(); |
| 84 | } |
| 85 | |
| 86 | DEFINE_NATIVE_ENTRY(RegExp_getIsMultiLine, 0, 1) { |
| 87 | const RegExp& regexp = RegExp::CheckedHandle(zone, ptr: arguments->NativeArgAt(index: 0)); |
| 88 | ASSERT(!regexp.IsNull()); |
| 89 | return Bool::Get(value: regexp.flags().IsMultiLine()).ptr(); |
| 90 | } |
| 91 | |
| 92 | DEFINE_NATIVE_ENTRY(RegExp_getIsUnicode, 0, 1) { |
| 93 | const RegExp& regexp = RegExp::CheckedHandle(zone, ptr: arguments->NativeArgAt(index: 0)); |
| 94 | ASSERT(!regexp.IsNull()); |
| 95 | return Bool::Get(value: regexp.flags().IsUnicode()).ptr(); |
| 96 | } |
| 97 | |
| 98 | DEFINE_NATIVE_ENTRY(RegExp_getIsDotAll, 0, 1) { |
| 99 | const RegExp& regexp = RegExp::CheckedHandle(zone, ptr: arguments->NativeArgAt(index: 0)); |
| 100 | ASSERT(!regexp.IsNull()); |
| 101 | return Bool::Get(value: regexp.flags().IsDotAll()).ptr(); |
| 102 | } |
| 103 | |
| 104 | DEFINE_NATIVE_ENTRY(RegExp_getIsCaseSensitive, 0, 1) { |
| 105 | const RegExp& regexp = RegExp::CheckedHandle(zone, ptr: arguments->NativeArgAt(index: 0)); |
| 106 | ASSERT(!regexp.IsNull()); |
| 107 | return Bool::Get(value: !regexp.flags().IgnoreCase()).ptr(); |
| 108 | } |
| 109 | |
| 110 | DEFINE_NATIVE_ENTRY(RegExp_getGroupCount, 0, 1) { |
| 111 | const RegExp& regexp = RegExp::CheckedHandle(zone, ptr: arguments->NativeArgAt(index: 0)); |
| 112 | ASSERT(!regexp.IsNull()); |
| 113 | if (regexp.is_initialized()) { |
| 114 | return Smi::New(value: regexp.num_bracket_expressions()); |
| 115 | } |
| 116 | const String& pattern = String::Handle(ptr: regexp.pattern()); |
| 117 | const String& errmsg = |
| 118 | String::Handle(ptr: String::New(cstr: "Regular expression is not initialized yet." )); |
| 119 | const String& message = String::Handle(ptr: String::Concat(str1: errmsg, str2: pattern)); |
| 120 | const Array& args = Array::Handle(ptr: Array::New(len: 1)); |
| 121 | args.SetAt(index: 0, value: message); |
| 122 | Exceptions::ThrowByType(type: Exceptions::kFormat, arguments: args); |
| 123 | return Object::null(); |
| 124 | } |
| 125 | |
| 126 | DEFINE_NATIVE_ENTRY(RegExp_getGroupNameMap, 0, 1) { |
| 127 | const RegExp& regexp = RegExp::CheckedHandle(zone, ptr: arguments->NativeArgAt(index: 0)); |
| 128 | ASSERT(!regexp.IsNull()); |
| 129 | if (regexp.is_initialized()) { |
| 130 | return regexp.capture_name_map(); |
| 131 | } |
| 132 | const String& pattern = String::Handle(ptr: regexp.pattern()); |
| 133 | const String& errmsg = String::Handle( |
| 134 | ptr: String::New(cstr: "Regular expression is not initialized yet. " )); |
| 135 | const String& message = String::Handle(ptr: String::Concat(str1: errmsg, str2: pattern)); |
| 136 | const Array& args = Array::Handle(ptr: Array::New(len: 1)); |
| 137 | args.SetAt(index: 0, value: message); |
| 138 | Exceptions::ThrowByType(type: Exceptions::kFormat, arguments: args); |
| 139 | return Object::null(); |
| 140 | } |
| 141 | |
| 142 | static ObjectPtr ExecuteMatch(Zone* zone, |
| 143 | NativeArguments* arguments, |
| 144 | bool sticky) { |
| 145 | const RegExp& regexp = RegExp::CheckedHandle(zone, ptr: arguments->NativeArgAt(index: 0)); |
| 146 | ASSERT(!regexp.IsNull()); |
| 147 | GET_NON_NULL_NATIVE_ARGUMENT(String, subject, arguments->NativeArgAt(1)); |
| 148 | GET_NON_NULL_NATIVE_ARGUMENT(Smi, start_index, arguments->NativeArgAt(2)); |
| 149 | |
| 150 | #if !defined(DART_PRECOMPILED_RUNTIME) |
| 151 | if (!FLAG_interpret_irregexp) { |
| 152 | return IRRegExpMacroAssembler::Execute(regexp, input: subject, start_offset: start_index, |
| 153 | /*sticky=*/sticky, zone); |
| 154 | } |
| 155 | #endif |
| 156 | return BytecodeRegExpMacroAssembler::Interpret(regexp, str: subject, start_index, |
| 157 | /*sticky=*/is_sticky: sticky, zone); |
| 158 | } |
| 159 | |
| 160 | DEFINE_NATIVE_ENTRY(RegExp_ExecuteMatch, 0, 3) { |
| 161 | // This function is intrinsified. See Intrinsifier::RegExp_ExecuteMatch. |
| 162 | return ExecuteMatch(zone, arguments, /*sticky=*/false); |
| 163 | } |
| 164 | |
| 165 | DEFINE_NATIVE_ENTRY(RegExp_ExecuteMatchSticky, 0, 3) { |
| 166 | // This function is intrinsified. See Intrinsifier::RegExp_ExecuteMatchSticky. |
| 167 | return ExecuteMatch(zone, arguments, /*sticky=*/true); |
| 168 | } |
| 169 | |
| 170 | } // namespace dart |
| 171 | |